Slashdot Mirror
Looking To Spammers To Solve Hard AI Problems
An anonymous reader writes "With bots getting closer to beating text-based CAPTCHAs for good, New Scientist points out that when they do, OCR technology will at least have advanced. The article goes on to suggest that whatever kind of reverse Turing Test that comes next should be chosen to motivate spammers to solve other pressing AI problems, such as image recognition. Are there any other problems that criminal crowdsourcing could help with?"
it has simply used existing OCR-type technology on a slightly (and I want to emphasize "slightly") different problem. Different character sets, if you will.
Spammers sell their code to other spammers all the time.
Facial recognition is not only pretty good, but is available in consumer applications. Google's Picasa does it quite well for your personal photos, and Face.com can go through your Facebook photos and quite accurately suggest tags.
Picasa is a popular image management program that has supported facial recognition since last year: http://www.techcrunch.com/2008/09/02/picasa-refresh-brings-facial-recognition/
I havnt used it, so im not sure how good it is.
Certainly not a full-on AI problem, just parameterize the flow density and flow rate and define a decent model and cost function, and run it through an NLP solver.
Except that it's really a discrete problem, with a solution that likely has sensitive dependence on initial conditions (i.e., chaotic), and would result in symptoms such as "bus bunching": http://en.wikipedia.org/wiki/Bus_bunching
We know where leadership by an anti-intellectual "strongman" who scapegoats minorities and likes boisterous rallies goes
CAPTCHAs have been dead for a long time already. Please direct me to the spam software that can actually read and interpret these for me, because I have about an 80% failure rate. I'm human, the very thing that is supposed to be able to figure all this out. If I see a site asking me to type in some obscure word or number, I click elsewhere. It's just too much trouble.
Spammers aren't using software to solve this problem anyway! Bold statement you might say? Maybe. Travel your backside to Asia, or, from the comfort of your own chair you could visit sites like sulit.com.ph (think craigs list wanna-be, it's that kind of thing) Every 3rd advert is asking for 'writers' that can log in to forums and post at least 3 or more messages before getting banned. How much does the lucky employee earn? About $200 USD and up per month. It's real money. So who is paying for this? People like the PHB in 1st world corporate wasteland, maybe your CEO thinking it's a good way to get more hits, maybe you. (No, not you personally) Evidently it works or the money wouldn't be flowing, and you wouldn't have 3000 people advertising this service each and every day.
Micro payments are terrible ideas, first because it violates basic net neutrality principles...
Methinks you have no idea what "net neutrality" actually means. What does paying to post on a forum have to do with net neutrality?
... and then they built the supercollider.
The new iPhoto '09 does a good job finding faces in clear pictures, and if you tell it where to look in fuzzy photos it learns and can find the face well there too. I have to agree facial recognition is making big steps forward. And no, it's not because I love my Apple that I mention this, it's because that's the latest facial recognition software I've used.
Hmmm, if you smoothed out the data so that it was say, averaged over an hour, and force it to be continuous, could you get something going then.
My point is that's about the worst assumption you can make.
We know where leadership by an anti-intellectual "strongman" who scapegoats minorities and likes boisterous rallies goes
You only have to get the word that OCR can recognize right. Just try guessing which of the two words OCR can't recognize and type some random gibberish instead of that word, it will let you through.
Converting the stereogram into a depth map: not very hard I think; at least, easier than for most humans. You look for repeating patterns along horizontal lines. Depending on whether the pattern repeats itself squeezed or stretched, it corresponds to negative or positive depth changes. The next problem is interpreting the depth map as an image to answer the captcha challenge (Q: what do you see here? A: shark), but it would be much more user-friendly to present the depth map directly to the user. I once read about the idea generate pictures from 3-dimensional models with arbitrary angles of view ("mother with child viewed from above"). The brain is much better at recognizing such pictures than computer vision software. A problem is that the web server needs to judge whether the answer given by the visitor is correct with a close-to-zero chance of guessing correct.
Avantslash: low-bandwidth mobile slashdot.
The general solution to this problem is for you to modify your software so that links in blog comments are served to add rel="nofollow" to all of the links. See http://en.wikipedia.org/wiki/Nofollow for more details. Of course that will not make the spam comment posts go away immediately but if the technique is rolled out widely, then the SEOs will figure out that posting spam blog comments does not gain them anything.
Actually this exists: http://spamornot.org/